Home Computer Audio Asylum

Music servers and other computer based digital audio technologies.

RE: Poor Clock architecture +1

"I would love to hear a DAC or streamer that Tony designed..since he says everything is broken or compromised. I would like to see what compromises he decides to make. Would certainly be interesting."

1. Streamer

Take that streamer and add a word clock input to it or a master clock input to it, and enable it to run in external clock mode. This is a simple change, and will allow the streamer to work well with DACs that export their clock. This is an extra cable with SPDIF, but no extra wires are needed with I2S, just changing the direction of the clock signal so that it goes from in the opposite direction (from the DAC to the streamer). Of course this requires a DAC that is compatible, i.e. runs I2S as the clock master. However this is already designed into the I2S specification, nothing new about it. This is the method that one box CD players have long used to get digital data from their transport section to their DAC section.

There is an inconvenience with this approach. Since the clock that determines sample rate is at the DAC it must run at the correct sample rate. There are several ways to deal with this:

1. User manually sets the sample rate at the DAC for each playlist that is at a specified sample rate. User must change this when playing albums at a different sample rate. (Inconvenience)

2. Transport or music server contains DSP that converts varying sample rates in individual files to a fixed sampling rate by upsampling. (More convenient, but sound quality now depends on DSP in the computer and issues arise as to 44.1x vs 48x family of sample rates.)

3. An additional control channel runs from the transport or server to the DAC that tells it what sample rate to run at. (This adds cost and software complexity as well as additional signal paths from the computer to the DAC that might couple noise.)

If the goal is the best possible sound quality approach (1) is preferred, or approach (2) can be used if upsampling at the transport is preferred because of filter issues, but unless the upsampling algorithms works as well at non-integer ratios as integer ratios it will still be necessary to manually switch the DAC to get best possible sound quality. This leads to a strategy of using (2) for casual listening and reserving the slight inconvenience of (1) for critical listening. Personally, I would be happy enough with a simple approach whereby the DAC runs at only one speed, e.g. DSD256. This greatly reduces the complexity of the DAC, removing most of the complex logic circuitry that creates digital noise and ensuring that all of the rest is running at very high rates far from the audio band.

It is possible for streamers to export I2S to the DAC with clock master at the streamer if the I2S cabling is very short and if the transport has a high quality clock that runs the output stream. However, this clock will have to be isolated from the circuitry that does any kind of packet processing, and the signals across the barrier have to have the same directionality as the correct clock architecture (data goes one way, clock goes the other). In effect, the "DAC" has been extended and the line between the DAC and the transport now is inside one box, just as it is in a one box CD player. Unfortunately, will add to the cost of the device, as isolators will be needed, separate ground and power planes, separate power supplies, separate clocks, etc. The entire product must be built of "audiophile grade" components. If the physical separation is done in a separate box at the DAC then only this box needs to be "audiophile grade" and the transport box can use commodity parts.

2. DAC

I would start with Miksa's DAC design, with the exception that I would replace the Amenero USB to I2S board with a simpler board that has an an isolated I2S master interface, a shift register/reclocker, a high quality master clock and the internal I2S interface to the rest of Miksa's circuitry. I would fix this DAC to run only at this one speed, all speed and format adaptation would have to be done at the source.

This will work best if the transport is an Ethernet to I2S converter and not USB, because this will allow use of a distant but high powered multi-core procesor needed to run the best quality sample rate conversions and other DSP at DSD256 speed.





Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar



This post is made possible by the generous support of people like you and our sponsors:
  Parts Connexion  


Follow Ups Full Thread
Follow Ups

FAQ

Post a Message!

Forgot Password?
Moniker (Username):
Password (Optional):
  Remember my Moniker & Password  (What's this?)    Eat Me
E-Mail (Optional):
Subject:
Message:   (Posts are subject to Content Rules)
Optional Link URL:
Optional Link Title:
Optional Image URL:
Upload Image:
E-mail Replies:  Automagically notify you when someone responds.